Design of Robust Controllers for Load Reduction in Wind Turbines

This thesis determines a design methodology of robust and multivariable controllers based on the H∞ norm reduction and on LPV (Linear Parameter Varying) techniques for load reduction in wind turbines. In order to do this, a 5 MW offshore wind turbine model based on the ‘Upwind’ European project is developed using GH Bladed, which is a wind turbine modelling specific software package. These controllers work in the above rated control zone, where the non-linearities of the wind turbine appear with more intensity. The main control objective in this zone is to keep the generator working at the nominal values of rotational speed and torque to correctly extract the nominal electric power in high winds. Furthermore, new control objectives are included to mitigate the loads in different components of the wind turbine, which involves the need of a multivariable control design. The family of linear models extracted from the non-linear model is used to design the proposed controllers. In this work, the family of linear models extracted from the GH Bladed is high ordered due to the complexity and accuracy of the wind turbine model. The Robust Control and LPVMAD MATLAB toolboxes are used to make the controller synthesis. LPVMAD is a toolbox developed by the scientific control group directed by Prof. Dr. Carsten Scherer at the Stuttgart University.

After an exhaustive analysis of the State of the Art about the wind turbine control systems, a baseline control strategy based on classical control methods is initially designed. Five monovariable, MISO (Multiple Input Single Output) and multivariable robust control strategies, based on the H∞ norm reduction, are presented to improve the benefits of the baseline controller. These controllers fulfill some control objectives to mitigate the loads in the wind turbine: generator speed regulation, drive train mode damping, tower first fore-aft and side-to-side first mode damping and rotor alignment. The designed H∞ controllers generate control signals of generator torque, collective pitch blade angle and individual pitch angles for each blade. On the other hand, two LPV control strategies are designed to improve the generator speed regulation in the above rated zone generating collective pitch angle set-point values. The first LPV controller consists of the interpolation of three H∞ controllers designed in three different operational points. The second LPV controller synthesis is based on a LMI (Linear Matrix Inequalities) solution using the LPVMAD toolbox and a wind turbine LPV model. The wind turbine multivariable LPV modelling process is also explained in this thesis.

The designed controllers are validated in GH Bladed and an exhaustive analysis is carried out to calculate the fatigue load reduction on the wind turbine components, as well as to analyze load mitigation in some extreme cases. The controllers are tested in a real time prototype which allows to carry out HIL (Hardware in the Loop) simulations. A GUI interface tool is developed in MATLAB to determine a sequential method making easier the controller design explained in this thesis. Finally, the proposed design methodology of robust and multivariable controllers is applied to a commercial 3 MW wind turbine.